资源环境科技发展态势分析平台

There is a substantial body of work that estimates residential water demand as a function of price, household characteristics, and weather. While most results suggest demand is inelastic, researchers have found qualitatively different estimates of price elasticity-both elastic (>1) and inelastic (<1). Inconsistency in the choice of the price signal, its instruments, and appropriate weather variables across studies makes it difficult to determine if the differences in elasticity estimates are real-due to location characteristics-or an artifice of model specification. Predicting accurate demand responses to price changes enables water utilities to stabilize and anticipate future revenue while meeting conservation goals. However, specifying residential water demand with the appropriate model (under block rate structure) is challenging for three reasons: (1) uncertainty around the appropriate price signal, (2) a lack of theoretical motivation for determining which (and how) weather variables affect demand, and (3) simultaneity issues in estimation require the use of instruments, which vary across studies. Therefore, to elucidate the effects of model choices on elasticity, we use a single location and systematically vary the specification of price, instruments, and weather variables across a suite of models. We reaffirm past results that households respond to the average price. We find reasonably stable and robust estimates of price elasticity across different weather and price instrument specification-with some notable exceptions-implying that low-cost information like average temperature is sufficient to accurately reflect the data-generating process, reducing costs and allowing for a parsimonious model.